Cognition

1. The Cognitive Revolution
1. Chomsky vs Skinner
2. The Failure of Associationism
3. Artificial Intelligence
4. Mental Images
2. Thinking
1. What is Thinking?
2. Mental Representations
3. Reasoning
3. Problem Solving
1. What is Problem Solving?
2. Trial & Error, Algorithms, and Heuristics
3. Tackling Problems
4. Problem Solving Traps
5. Creativity
4. Language
1. What is Language?
2. Linguistics
3. Grammar and Syntax
4. Development and Acquisition of Language

1. The Cognitive Revolution
(top)
• From about 1920 until the 1960’s, academic psychology was dominated by Behaviorism.
• Influential figures included Watson (the child abuser) and Skinner.
• The rise of Behaviorism was a reaction to the subjectivity that was common in the late 1800’s and early 1900’s.
• Behaviorism simplified experimental psychology by restricting the field to objectively observable behaviors.
• Behaviorist approach also facilitates the use of animal analogues since no one worries about what a rat or pigeon might be thinking.
• Recent critics have pointed out that the experimental set-ups used by Behaviorists are over-simplified: they lack natural stimuli that are part of the normal environment.
• Perhaps Skinner could teach pigeons to play the piano, but that was a highly artificial behavior learned under extremely artificial conditions.
• As a result, the process teaches us little about problem solving by pigeons or humans.
• Skinner proposed that psychology would make great strides as a science once it was put on an objective, ie behavioral, footing. He predicted the rise of a "technology of behavior."
• Skinnerian principles have had successful application in psychotherapy, toilet training, and control of institutional populations.
• Most psychological phenomena have not been well explained within the behavioral framework.
• In the 1950’s and 60’s a number of factors led to the "cognitive revolution," a return to the concept of a structured and functioning mental life underlying psychology.
1. Chomsky versus Skinner
(top)
• In 1957, Skinner published a book titled Verbal Behavior in which he analyzed human speech from a behaviorist viewpoint.
• In 1959, MIT linguist Noam Chomsky (probably the world's most famous linguist) wrote a critical review of Skinner's book that slammed his theories.
• His criticisms were of three types:
• The rules worked out for rats in Skinner boxes cannot be clearly applied to humans in their real world
• When a rat pushes a lever in response to a light coming on, we say that the response, the lever press, is under the control of a stimulus, the light
• But when a person makes a statement in response to something such as a painting, it is not clear what the controlling stimulus actually is
• There are thousands of appropriate statements in a given situation--we can try to pick out some controlling stimulus once we have heard the statement, but it will not be some obvious objective feature of the environment (like a light)
• The precision used in laboratory studies was missing from an analysis of real human speech
• The idea of reinforcement became so vague as to be meaningless
• Skinner stretched the meaning of the word behavior to the point of nonsense and even used his nonsensical meaning inconsistently
• In the 1960's, Chomsky continued to contradict the then prevailing view that children learn simply by imitating their parents.
• He based his proposal on analysis of the spoken language of young children and of languages from around the world.
• He proposed a universal grammar underlying all human languages.
• He found that children have built in grammatical ability that increases as they mature.
• They begin to make certain types of sentences regardless of what language they are learning.
• They make predictable errors ("We goed to the store last night").
• Chomsky proposed that the brain contains structures he called the language-acquisition device.
• It appears that different brain areas are active in the learning of a second language (as opposed to one's native language).
2. The Failure of Associationism
(top)
• Behaviorism proposed that the mind is unimportant, that there is nothing to be gained by supposing that somehting in the mind influences behavior in a way that cannot be observed directly by connecting stimulus with response.
• Within Behaviorism, learning (that is, changes in behavior) is the result of association.
• The association of ideas is a theory that goes back to eighteenth century British philosophers John Locke, David Hume, and others.
• Locke is famous the notion that the mind of an infant is a tabula rasa, or blank slate, on which experience can write (sounds like Watson’s 12 healthy infants).
• Under this theory, though is governed by two laws: contiguity and resemblance.
• Contiguity holds that things experienced together become associated.
• Resemblance holds that whatever is associated with a thing automatically becomes associated with similar things.
• The behaviorists sometimes believed that the properties of conditioning were empirical support for associationism.
• Associationism does not work as a model for how we think.
• People distinguish individual things from the properties with with they are associated.
• Even very young children know that there is a difference between their toy and another toy even if the other toy is apparently identical to the first.
• Associationism also has trouble with our ability to see the meaning of how things are arranged.
• This is obvious in language: "The dog bit the boy" and "The boy bit the dog" are clearly very different in meaning even though the pieces of the stimuli are identical.
• When we interpret these sentences, we are not simply lumping together the ideas we have associated with dogness, boyness and biting.
• The notion of an empty mind also runs into the frame problem. If our minds are truly blank to start with, how could we associate anything with a dog?
• What is a dog anyway? On any occasion where we could have opportunity to associate, say, odor with a dog, we are experiencing a particular dog not the abstract class of "dogness."
• You could argue that with repeated exposure to individual dogs, a series of associations is made that then defines the abstract entity.
• This is perhaps not quite circular, you could imagine the mind doing a sort of factor analysis on all its experience to date.
• The huge number of possible associations, both with things that are generally agreed to exist and with phantom things, would insure that this process would require a tremendous amount of time—probably longer than you have lived.
• Another way of looking at this is to consider the gesture of pointing. Even young children know that when you point at Spot and say "dog" you are refering to the four-legged smelly thing and not the collar on him, his tail, the dog and the table he’s standing on, the left side of the room, etc.
3. Artificial intelligence
(top)
• The rise in the use of computers in the 50’s and 60’s influenced the return of the mind to psychology in two ways.
• The computer was a machine that appeared to think. Information processing was examined as an abstract entity that could be embodied in an electronic device or perhaps in a brain.
• Information processing could be examined on paper. Once a hypothetical mechanism was proposed, researchers could try to observe it in humans or model it in computers.
• This process is rigorous and objective and thus has the supposed benefits of the behavioral approach.
• Artificial intelligence research showed that the most ordinary of human behaviors requires a high degree of mental complexity.
• "Common sense" is perhaps less than common among humans, but it is virtually unheard of in computers.
• If Behaviorism were correct, then one would suppose that it would be easy to program a computer to simulate human or rat behavior. The rules of conditioning are simple, implementing them is not.
4. Mental Images
(top)
• Think of a number from one to ten.
• Does that number have any curved shapes in it or only straight lines?
• To answer the question, you formed a mental image of the number.
• When we think of something, we often "see" it.
• Or hear, taste, feel, smell….
• A mental image is like a sense image in the absence of a stimulus.
• Mental images are real.
• Stephen Kosslyn et al did experiments in the 1970s in which people had to find things on maps.
• Subjects were asked to focus on one spot on the map and then asked to find another thing. The farther away it was, the longer it took subjects to find it.
• The interesting thing is that the map was shown to the subjects, then taken away before the search task.
• It seemed as thought the subjects were moving their mind's eyes over the mental image of the map.
• Similarly, when subjects were asked to compare two objects which have been rotated to different orientations, the greater the rotation, the longer it takes to say whether the two objects are the same.
• Brain scans show similar patterns of activation when a subject is thinking of something and when he or she is actually looking at it or doing it.
• Eg, people remembering a melody show temporal lobe activity as do those listening to a melody.
• Note that mental images depend on memory and are therefore subject to its errors, distortions and other failures.
2. Thinking
(top)
1. What is Thinking
(top)
• Thinking is the manipulation of representations of information in the mind
• permits us to draw inferences and make connections (ie, reason)
• active processes
• often (usually?) goal-oriented
• Based on the work of Wickens and others, Bernstein describes a "circle of thought" (p 250-251). To me, this circle is reminiscient of the scientific method.
• Cognition is the general term for all mental activities connected to acquiring, processing, retaining and applying knowledge
• Sensation and perception, learning, and memory can be classified as cognitive activities
• The raw material of thinking consists of at least five types of entities: cognitive maps, concepts, mental images, narratives, and propositions.
2. Mental Representations
(top)
• Cognitive maps
• Remember Tolman’s rats? People do the same thing: we store information about spatial relationships between things.
• Concepts
• Is a dog an animal?
• It is tempting to imagine that all of our thinking is based on mental images, but this is almost certainly not the case.
• We can call up an image of a dog and an image of an animal and see if they are the same. This clearly will not work.
• Instead, we use another type of mental entity, a concept.
• A concept is a mental construction that permits us to group things together; a mental category.
• We can define rules to determine whether or not something belongs to a given category. In this case we have a formal concept.
• If something follows the rules, we say that it has the attributes of the category, or that it fits the concept.
• Some things seem to go together by themselves. Eleanor Rosch proposed that our everyday experience leads us to form natural concepts.
• These are more loosely defined; we may not be explicitly aware of the rules.
• Some things fit more readily to a concept than others. Rosch proposed that the best or most typical fit is a prototype of the concept. Things which are dissimilar to the prototype are harder to put in the concept.
• For example, we readily call an apple or a banana a fruit. But we hesitate with a tomato or a zucchini and balk altogether at an olive or a coconut.
• Arguably, olive fits the formal concept of fruit but not the natural concept.
• Concepts can be organized into hierarchies.
• There are concepts and subconcepts and sub-subconcepts, etc.
• Eg, furniture, chair, rocking chair…
• The ability to form concepts changes as children develop. We will look at Piaget's stages of development in a later week.
• Mental images were discussed above.
• Propositions
• We form ideas about the relationship between concepts. We can state these relationships as propositions.
• Humans have a highly developed ability to manipulate propositions. We call this ability reasoning or formal reasoning. This is discussed further below.
• Narrative
• Sometimes knowledge is not spatial, is not a category, is not visual, and does not contain explicit relationships between things. The knowledge is laid out in the same way that our experience of an event is.
• The contents of our episodic memory can be recalled as stories. Stories can be very rich in knowledge, though we usually need to do some work to dig out specific pieces of information.
3. Reasoning
(top)
• Reasoning is a form of thinking by which we see connections between things. You could picture this as the manipulation of propositions.
• In the strictest sense, reasoning is a process of determining whether a particular conclusion is supported by a set of premises.
• Earlier in the semester, I mentioned that science depends on deductive and inductive reasoning.
• Deductive reasoning draws a conclusion from a set of premises. If the premises are true the conclusion must be true.
• Deductive reasoning can be summarized by a set of rules (a computer can easily be taught to do it perfectly).
• Inductive reasoning requires a leap of inference. A conclusion is deemed to be likely, based on a set of premises.
• This is much harder to summarize in rules (and is quite difficult to teach to a computer).
• Diagnostic reasoning used both deduction and induction as well as a sort of mental experimentation.
• This is what a doctor or auto mechanic uses to figure out what is wrong with your body or car, or what a detective uses to solve a crime.
• Different conclusions are dreamed up and then compared with the premises.
• The one that fits best is considered to be the proper diagnosis.
• People attempt to summarize this process with rules (called heuristics) and procedures (called algorithms), but it is subtle and difficult to summarize. There has been some success in programming this type of reasoning into computers.
3. Problem Solving
(top)
1. What is Problem Solving?
(top)
• Problem solving is the mental activity and outward behavior directed toward reaching some goal.
• We define a problem as a situation in which set of circumstances is different from present circumstances and not immediately attainable.
• To determine the procedure to get from here to there, and perhaps to carry out that procedure, is to solve the problem.
• By this broad definition, we face a large number of problems every day, many of which we solve without even being aware of the problem.
• In some cases, we are definitely conscious of a problem, devoting mental resources to its solution, and perhaps experiencing some anxiety until it is solved.
• There are processes that we regularly use when faced with a problem.
• We can learn techniques that can be used to improve our problem solving ability.
2. Trial & Error, Algorithms, and Heuristics
(top)
• A popular, easy to learn and often-successful technique is trial and error: guessing at a solution, trying it and repeating until the problem is solved.
• This does not necessarily require much preparation or intelligence. (Arguably Thorndike's cats were using this to get out of their puzzle boxes.)
• But, trial and error is a bad choice if
• the number of possible solutions is large (I wouldn't suggest this technique to learn someone's phone number).
• some of the possible solutions are harmful (you wouldn't want to learn to drive on the highway by trial and error).
• there is a possibility that the problem has no solution.
• When all else fails, follow the directions: An algorithm is a set of steps that are followed in order to reach a certain solution.
• For example: "Wet hair, apply shampoo, lather, rinse and repeat."
• Computer programs are examples of algorithms.
• Algorithms are helpful because they often do not require a high degree of expertise to be successful.
• Consider CPR: You do not have to go to medical school for umpteen years to save someone's life, as long as you follow the directions.
• There was optimism at one time that we could write algorithms for everything that expensive experts are called on to do (for example, we could program computers to diagnose illnesses and save on our health insurance premiums).
• Unfortunately, this has not worked. Experts often do not use algorithms, even when they describe their procedures as a series of steps.
• Incidentally, this is partly a matter of concepts and is analogous to chunking in short term memory.
• A good chess player does not see individual pieces and possible single moves, instead he or she has concepts that correspond to patterns of many pieces and multiple possible moves.
• An algorithm may be guaranteed to work, but is not necessarily the best approach since a solution may many steps or many repetitions to achieve (of course that's ok if you are a computer who can do millions or billions of steps every second).
• Consider the phone number problem.
• A heuristic is a shortcut or rule-of-thumb that might quickly bring about a solution, but is not guaranteed to solve the problem each time.
• A heuristic to observe when faced with the phone number problem is "use the phone book."
• You won't necessarily find the number, but you will save a lot of time over starting at 000-0000.
• A heuristic approach is not static, and in fact contains an element of trial and error.
• The difference is that when an approach works, you store it away for future use.
• So-called expert systems on computers often work this way.
• They start with a set of things to try and then learn from their mistakes.
• There are some general heuristics that can be applied to any problem.
• Break the problem down into smaller sub-problems.
• Start at the beginning and the end and work toward the middle.
• We will consider more strategies in just a minute.
• Insight and intuition can dramatically reduce the time required to reach a solution.
• An insight is a sudden realization of a potential solution.
• Sometimes this occurs while we are mulling over the problem, but it often seems to come out of the blue, sometimes even after we have given up on a problem, at least for the moment. In any case, we do not seem to be able to force ourselves to have insight.
• This phenomenon has lead psychologists to propose the existence of non-conscious processes in problem solving.
3. Tackling Problems
(top)
• Psychologists Bransford and Stein have proposed an IDEAL approach to problem solving:
• Identify the problem.
• Define the problem and make a model of it to work with (eg, draw a picture).
• Explore and experiment with possible solutions.
• Act on the solution that seems best.
• Look back and evaluate whether the solution did what you wanted.
4. Problem Solving Traps
(top)
• Two common obstacles to problem solving are known as functional fixedness and mental set.
• Functional fixedness refers to our tendency to see things only in terms of their usual actions.
• Mental set refers our inclination to try to reuse solutions that have worked in the past.
• Once you have solved a certain type of problem, you assume that the next problem can be solved in the same way.
5. Creativity
(top)
• Creativity is the ability to come up with new or different information or new or different uses of old information.
• This can be very helpful in solving certain problems, but is not a help in all cases.
• Standard problem solving is based on convergent thinking, the use standard of information, reason and knowledge.
• Creative solutions are promoted by divergent thinking, which produces unconventional uses for knowledge or information.
• Creativity is elusive, but a number of psychologists (including Feldman) believe it can be promoted by certain practices:
• Take risks: consider oddball solutions
• Play devil's advocate: consider the opposite of the standard solution.
• Redefine the problem.
• Challenge assumptions.
• Sometimes creativity is described as a "right-brained" ability, because the right cerebral hemisphere has been found to be better with holistic, intuitive, imaginative and non-verbal tasks.
• But Ned Herrman, a psychologist formerly with General Electric, points out that real world creativity requires not just the ability to generate fluffy and fun ideas, but the means to test and adapt these to the constraints of reality.
• He has proposed a whole brain model for creativity in which the left and right cerebral hemispheres and the left and right limbic systems contribute to the creative process
4. Language
(top)
1. What is Language?
• Language is the use of symbols to represent concrete and abstract aspects of the world.
• In a full-fledged language, the symbols are largely arbitrary (that is, there is no relationship between form and meaning).
• Nosing the can of cat food when you are hungry does not qualify as using language, but nosing it to mean "290 shopping days until Christmas" might.
• Some words do sound like their meanings (eg, "burp"), a phenomenon known as onomatopoeia.
• Likewise, written language may contain pictograms, stylized pictures of the things meant by the words.
• The number of possible utterances is infinite (linguists say that language is generative).
• Language is shared, and so can be used as a means of communication.
• True language can be used to express ideas about things that are distant in space and/or time (linguists say that language has the property of displacement).
• Human language is intimately tied to human cognition. No doubt we have thoughts which are not expressed in language, but if we examine the contents of our thoughts, we find them full of language.
• Similarly, it is hard to imagine how we would communicate complex ideas without using language.
2. Linguistics
(top)
• Linguistics is the study of the structure, development and use of language.
3. Grammar and Semantics
(top)
• Every language uses symbols (aka words).
• Semantics refers to the meanings of the symbols.
• Semantics also refers to meaning in general (ie, the meaning of sentences).
• Languages have a syntax or grammar.
• Syntax, or grammar, is the set of rules that determine how the symbols (words) can be put together.
• Both semantics and syntax are needed for complex utterances.
• These interact to determine meaning.
4. Development & Acquisition of Language
(top)
• Is language an innate ability in humans?
• What do we mean by "innate"?
• Language is like many other types of behaviors in that it requires an interaction of environment (parents, etc) with intrinsic factors within the developing brain (or mind) of an infant.
• An oversimplification contrasts innate theories with learning theories.
• The pure Skinnerian approach was long ago discredited, though there may be some operant conditioning at work: if children are reinforced for speaking correctly, the frequency of correct speech might therefore increase. Presumably there would be some shaping involved.
• Children who are spoken to more, do speak more at an earlier age, but there is a big frame problem here: what is the controlling stimulus?
• Steven Pinker put the innatist view bluntly: "People know how to talk in more or less the sense that spiders know how to spin webs."
• There is still some resistance to this point of view.
• There may be legitimate disagreement over how much of language knowledge is acquired from the environment and how much is innately available to the developing brain.
• Clearly both environmental input and a developing human child are required for normal speech to come about; but the same environmental input will not allow a chimpanzee to use language normally. In fact, it will not allow an adult to learn a second language normally.

Anthony G Benoit  abenoit@trcc.commnet.edu
(860) 885-2386

Revised